lib/Geocode.php

   1 <?php
   2
   3 namespace Nominatim;
   4
   5 require_once(CONST_BasePath.'/lib/PlaceLookup.php');
   6 require_once(CONST_BasePath.'/lib/Phrase.php');
   7 require_once(CONST_BasePath.'/lib/ReverseGeocode.php');
   8 require_once(CONST_BasePath.'/lib/SearchDescription.php');
   9 require_once(CONST_BasePath.'/lib/SearchContext.php');
  10
  11 class Geocode
  12 {
  13     protected $oDB;
  14
  15     protected $aLangPrefOrder = array();
  16
  17     protected $bIncludeAddressDetails = false;
  18     protected $bIncludeExtraTags = false;
  19     protected $bIncludeNameDetails = false;
  20
  21     protected $bIncludePolygonAsPoints = false;
  22     protected $bIncludePolygonAsText = false;
  23     protected $bIncludePolygonAsGeoJSON = false;
  24     protected $bIncludePolygonAsKML = false;
  25     protected $bIncludePolygonAsSVG = false;
  26     protected $fPolygonSimplificationThreshold = 0.0;
  27
  28     protected $aExcludePlaceIDs = array();
  29     protected $bDeDupe = true;
  30     protected $bReverseInPlan = false;
  31
  32     protected $iLimit = 20;
  33     protected $iFinalLimit = 10;
  34     protected $iOffset = 0;
  35     protected $bFallback = false;
  36
  37     protected $aCountryCodes = false;
  38
  39     protected $bBoundedSearch = false;
  40     protected $aViewBox = false;
  41     protected $aRoutePoints = false;
  42     protected $aRouteWidth = false;
  43
  44     protected $iMaxRank = 20;
  45     protected $iMinAddressRank = 0;
  46     protected $iMaxAddressRank = 30;
  47     protected $aAddressRankList = array();
  48     protected $exactMatchCache = array();
  49
  50     protected $sAllowedTypesSQLList = false;
  51
  52     protected $sQuery = false;
  53     protected $aStructuredQuery = false;
  54
  55     protected $oNormalizer = null;
  56
  57
  58     public function __construct(&$oDB)
  59     {
  60         $this->oDB =& $oDB;
  61         $this->oNormalizer = \Transliterator::createFromRules(CONST_Term_Normalization_Rules);
  62     }
  63
  64     private function normTerm($sTerm)
  65     {
  66         if ($this->oNormalizer === null) {
  67             return $sTerm;
  68         }
  69
  70         return $this->oNormalizer->transliterate($sTerm);
  71     }
  72
  73     public function setReverseInPlan($bReverse)
  74     {
  75         $this->bReverseInPlan = $bReverse;
  76     }
  77
  78     public function setLanguagePreference($aLangPref)
  79     {
  80         $this->aLangPrefOrder = $aLangPref;
  81     }
  82
  83     public function getMoreUrlParams()
  84     {
  85         if ($this->aStructuredQuery) {
  86             $aParams = $this->aStructuredQuery;
  87         } else {
  88             $aParams = array('q' => $this->sQuery);
  89         }
  90
  91         if ($this->aExcludePlaceIDs) {
  92             $aParams['exclude_place_ids'] = implode(',', $this->aExcludePlaceIDs);
  93         }
  94
  95         if ($this->bIncludeAddressDetails) $aParams['addressdetails'] = '1';
  96         if ($this->bIncludeExtraTags) $aParams['extratags'] = '1';
  97         if ($this->bIncludeNameDetails) $aParams['namedetails'] = '1';
  98
  99         if ($this->bIncludePolygonAsPoints) $aParams['polygon'] = '1';
 100         if ($this->bIncludePolygonAsText) $aParams['polygon_text'] = '1';
 101         if ($this->bIncludePolygonAsGeoJSON) $aParams['polygon_geojson'] = '1';
 102         if ($this->bIncludePolygonAsKML) $aParams['polygon_kml'] = '1';
 103         if ($this->bIncludePolygonAsSVG) $aParams['polygon_svg'] = '1';
 104
 105         if ($this->fPolygonSimplificationThreshold > 0.0) {
 106             $aParams['polygon_threshold'] = $this->fPolygonSimplificationThreshold;
 107         }
 108
 109         if ($this->bBoundedSearch) $aParams['bounded'] = '1';
 110         if (!$this->bDeDupe) $aParams['dedupe'] = '0';
 111
 112         if ($this->aCountryCodes) {
 113             $aParams['countrycodes'] = implode(',', $this->aCountryCodes);
 114         }
 115
 116         if ($this->aViewBox) {
 117             $aParams['viewbox'] = $this->aViewBox[0].','.$this->aViewBox[3]
 118                                   .','.$this->aViewBox[2].','.$this->aViewBox[1];
 119         }
 120
 121         return $aParams;
 122     }
 123
 124     public function setIncludePolygonAsPoints($b = true)
 125     {
 126         $this->bIncludePolygonAsPoints = $b;
 127     }
 128
 129     public function setIncludePolygonAsText($b = true)
 130     {
 131         $this->bIncludePolygonAsText = $b;
 132     }
 133
 134     public function setIncludePolygonAsGeoJSON($b = true)
 135     {
 136         $this->bIncludePolygonAsGeoJSON = $b;
 137     }
 138
 139     public function setIncludePolygonAsKML($b = true)
 140     {
 141         $this->bIncludePolygonAsKML = $b;
 142     }
 143
 144     public function setIncludePolygonAsSVG($b = true)
 145     {
 146         $this->bIncludePolygonAsSVG = $b;
 147     }
 148
 149     public function setPolygonSimplificationThreshold($f)
 150     {
 151         $this->fPolygonSimplificationThreshold = $f;
 152     }
 153
 154     public function setLimit($iLimit = 10)
 155     {
 156         if ($iLimit > 50) $iLimit = 50;
 157         if ($iLimit < 1) $iLimit = 1;
 158
 159         $this->iFinalLimit = $iLimit;
 160         $this->iLimit = $iLimit + min($iLimit, 10);
 161     }
 162
 163     public function setFeatureType($sFeatureType)
 164     {
 165         switch ($sFeatureType) {
 166             case 'country':
 167                 $this->setRankRange(4, 4);
 168                 break;
 169             case 'state':
 170                 $this->setRankRange(8, 8);
 171                 break;
 172             case 'city':
 173                 $this->setRankRange(14, 16);
 174                 break;
 175             case 'settlement':
 176                 $this->setRankRange(8, 20);
 177                 break;
 178         }
 179     }
 180
 181     public function setRankRange($iMin, $iMax)
 182     {
 183         $this->iMinAddressRank = $iMin;
 184         $this->iMaxAddressRank = $iMax;
 185     }
 186
 187     public function setViewbox($aViewbox)
 188     {
 189         $this->aViewBox = array_map('floatval', $aViewbox);
 190
 191         $this->aViewBox[0] = max(-180.0, min(180, $this->aViewBox[0]));
 192         $this->aViewBox[1] = max(-90.0, min(90, $this->aViewBox[1]));
 193         $this->aViewBox[2] = max(-180.0, min(180, $this->aViewBox[2]));
 194         $this->aViewBox[3] = max(-90.0, min(90, $this->aViewBox[3]));
 195
 196         if (abs($this->aViewBox[0] - $this->aViewBox[2]) < 0.000000001
 197             || abs($this->aViewBox[1] - $this->aViewBox[3]) < 0.000000001
 198         ) {
 199             userError("Bad parameter 'viewbox'. Not a box.");
 200         }
 201     }
 202
 203     public function setQuery($sQueryString)
 204     {
 205         $this->sQuery = $sQueryString;
 206         $this->aStructuredQuery = false;
 207     }
 208
 209     public function getQueryString()
 210     {
 211         return $this->sQuery;
 212     }
 213
 214
 215     public function loadParamArray($oParams)
 216     {
 217         $this->bIncludeAddressDetails
 218          = $oParams->getBool('addressdetails', $this->bIncludeAddressDetails);
 219         $this->bIncludeExtraTags
 220          = $oParams->getBool('extratags', $this->bIncludeExtraTags);
 221         $this->bIncludeNameDetails
 222          = $oParams->getBool('namedetails', $this->bIncludeNameDetails);
 223
 224         $this->bBoundedSearch = $oParams->getBool('bounded', $this->bBoundedSearch);
 225         $this->bDeDupe = $oParams->getBool('dedupe', $this->bDeDupe);
 226
 227         $this->setLimit($oParams->getInt('limit', $this->iFinalLimit));
 228         $this->iOffset = $oParams->getInt('offset', $this->iOffset);
 229
 230         $this->bFallback = $oParams->getBool('fallback', $this->bFallback);
 231
 232         // List of excluded Place IDs - used for more acurate pageing
 233         $sExcluded = $oParams->getStringList('exclude_place_ids');
 234         if ($sExcluded) {
 235             foreach ($sExcluded as $iExcludedPlaceID) {
 236                 $iExcludedPlaceID = (int)$iExcludedPlaceID;
 237                 if ($iExcludedPlaceID)
 238                     $aExcludePlaceIDs[$iExcludedPlaceID] = $iExcludedPlaceID;
 239             }
 240
 241             if (isset($aExcludePlaceIDs))
 242                 $this->aExcludePlaceIDs = $aExcludePlaceIDs;
 243         }
 244
 245         // Only certain ranks of feature
 246         $sFeatureType = $oParams->getString('featureType');
 247         if (!$sFeatureType) $sFeatureType = $oParams->getString('featuretype');
 248         if ($sFeatureType) $this->setFeatureType($sFeatureType);
 249
 250         // Country code list
 251         $sCountries = $oParams->getStringList('countrycodes');
 252         if ($sCountries) {
 253             foreach ($sCountries as $sCountryCode) {
 254                 if (preg_match('/^[a-zA-Z][a-zA-Z]$/', $sCountryCode)) {
 255                     $aCountries[] = strtolower($sCountryCode);
 256                 }
 257             }
 258             if (isset($aCountries))
 259                 $this->aCountryCodes = $aCountries;
 260         }
 261
 262         $aViewbox = $oParams->getStringList('viewboxlbrt');
 263         if ($aViewbox) {
 264             if (count($aViewbox) != 4) {
 265                 userError("Bad parmater 'viewboxlbrt'. Expected 4 coordinates.");
 266             }
 267             $this->setViewbox($aViewbox);
 268         } else {
 269             $aViewbox = $oParams->getStringList('viewbox');
 270             if ($aViewbox) {
 271                 if (count($aViewbox) != 4) {
 272                     userError("Bad parmater 'viewbox'. Expected 4 coordinates.");
 273                 }
 274                 $this->setViewBox($aViewbox);
 275             } else {
 276                 $aRoute = $oParams->getStringList('route');
 277                 $fRouteWidth = $oParams->getFloat('routewidth');
 278                 if ($aRoute && $fRouteWidth) {
 279                     $this->aRoutePoints = $aRoute;
 280                     $this->aRouteWidth = $fRouteWidth;
 281                 }
 282             }
 283         }
 284     }
 285
 286     public function setQueryFromParams($oParams)
 287     {
 288         // Search query
 289         $sQuery = $oParams->getString('q');
 290         if (!$sQuery) {
 291             $this->setStructuredQuery(
 292                 $oParams->getString('amenity'),
 293                 $oParams->getString('street'),
 294                 $oParams->getString('city'),
 295                 $oParams->getString('county'),
 296                 $oParams->getString('state'),
 297                 $oParams->getString('country'),
 298                 $oParams->getString('postalcode')
 299             );
 300             $this->setReverseInPlan(false);
 301         } else {
 302             $this->setQuery($sQuery);
 303         }
 304     }
 305
 306     public function loadStructuredAddressElement($sValue, $sKey, $iNewMinAddressRank, $iNewMaxAddressRank, $aItemListValues)
 307     {
 308         $sValue = trim($sValue);
 309         if (!$sValue) return false;
 310         $this->aStructuredQuery[$sKey] = $sValue;
 311         if ($this->iMinAddressRank == 0 && $this->iMaxAddressRank == 30) {
 312             $this->iMinAddressRank = $iNewMinAddressRank;
 313             $this->iMaxAddressRank = $iNewMaxAddressRank;
 314         }
 315         if ($aItemListValues) $this->aAddressRankList = array_merge($this->aAddressRankList, $aItemListValues);
 316         return true;
 317     }
 318
 319     public function setStructuredQuery($sAmenity = false, $sStreet = false, $sCity = false, $sCounty = false, $sState = false, $sCountry = false, $sPostalCode = false)
 320     {
 321         $this->sQuery = false;
 322
 323         // Reset
 324         $this->iMinAddressRank = 0;
 325         $this->iMaxAddressRank = 30;
 326         $this->aAddressRankList = array();
 327
 328         $this->aStructuredQuery = array();
 329         $this->sAllowedTypesSQLList = false;
 330
 331         $this->loadStructuredAddressElement($sAmenity, 'amenity', 26, 30, false);
 332         $this->loadStructuredAddressElement($sStreet, 'street', 26, 30, false);
 333         $this->loadStructuredAddressElement($sCity, 'city', 14, 24, false);
 334         $this->loadStructuredAddressElement($sCounty, 'county', 9, 13, false);
 335         $this->loadStructuredAddressElement($sState, 'state', 8, 8, false);
 336         $this->loadStructuredAddressElement($sPostalCode, 'postalcode', 5, 11, array(5, 11));
 337         $this->loadStructuredAddressElement($sCountry, 'country', 4, 4, false);
 338
 339         if (sizeof($this->aStructuredQuery) > 0) {
 340             $this->sQuery = join(', ', $this->aStructuredQuery);
 341             if ($this->iMaxAddressRank < 30) {
 342                 $this->sAllowedTypesSQLList = '(\'place\',\'boundary\')';
 343             }
 344         }
 345     }
 346
 347     public function fallbackStructuredQuery()
 348     {
 349         if (!$this->aStructuredQuery) return false;
 350
 351         $aParams = $this->aStructuredQuery;
 352
 353         if (sizeof($aParams) == 1) return false;
 354
 355         $aOrderToFallback = array('postalcode', 'street', 'city', 'county', 'state');
 356
 357         foreach ($aOrderToFallback as $sType) {
 358             if (isset($aParams[$sType])) {
 359                 unset($aParams[$sType]);
 360                 $this->setStructuredQuery(@$aParams['amenity'], @$aParams['street'], @$aParams['city'], @$aParams['county'], @$aParams['state'], @$aParams['country'], @$aParams['postalcode']);
 361                 return true;
 362             }
 363         }
 364
 365         return false;
 366     }
 367
 368     public function getDetails($aPlaceIDs, $oCtx)
 369     {
 370         //$aPlaceIDs is an array with key: placeID and value: tiger-housenumber, if found, else -1
 371         if (sizeof($aPlaceIDs) == 0) return array();
 372
 373         $sLanguagePrefArraySQL = getArraySQL(
 374             array_map("getDBQuoted", $this->aLangPrefOrder)
 375         );
 376
 377         // Get the details for display (is this a redundant extra step?)
 378         $sPlaceIDs = join(',', array_keys($aPlaceIDs));
 379
 380         $sImportanceSQL = $oCtx->viewboxImportanceSQL('ST_Collect(centroid)');
 381         $sImportanceSQLGeom = $oCtx->viewboxImportanceSQL('geometry');
 382
 383         $sSQL  = "SELECT ";
 384         $sSQL .= "    osm_type,";
 385         $sSQL .= "    osm_id,";
 386         $sSQL .= "    class,";
 387         $sSQL .= "    type,";
 388         $sSQL .= "    admin_level,";
 389         $sSQL .= "    rank_search,";
 390         $sSQL .= "    rank_address,";
 391         $sSQL .= "    min(place_id) AS place_id, ";
 392         $sSQL .= "    min(parent_place_id) AS parent_place_id, ";
 393         $sSQL .= "    country_code, ";
 394         $sSQL .= "    get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) AS langaddress,";
 395         $sSQL .= "    get_name_by_language(name, $sLanguagePrefArraySQL) AS placename,";
 396         $sSQL .= "    get_name_by_language(name, ARRAY['ref']) AS ref,";
 397         if ($this->bIncludeExtraTags) $sSQL .= "hstore_to_json(extratags)::text AS extra,";
 398         if ($this->bIncludeNameDetails) $sSQL .= "hstore_to_json(name)::text AS names,";
 399         $sSQL .= "    avg(ST_X(centroid)) AS lon, ";
 400         $sSQL .= "    avg(ST_Y(centroid)) AS lat, ";
 401         $sSQL .= "    COALESCE(importance,0.75-(rank_search::float/40)) $sImportanceSQL AS importance, ";
 402         if ($oCtx->hasNearPoint()) {
 403             $sSQL .= $oCtx->distanceSQL('ST_Collect(centroid)')." AS addressimportance,";
 404         } else {
 405             $sSQL .= "    ( ";
 406             $sSQL .= "       SELECT max(p.importance*(p.rank_address+2))";
 407             $sSQL .= "       FROM ";
 408             $sSQL .= "         place_addressline s, ";
 409             $sSQL .= "         placex p";
 410             $sSQL .= "       WHERE s.place_id = min(CASE WHEN placex.rank_search < 28 THEN placex.place_id ELSE placex.parent_place_id END)";
 411             $sSQL .= "         AND p.place_id = s.address_place_id ";
 412             $sSQL .= "         AND s.isaddress ";
 413             $sSQL .= "         AND p.importance is not null ";
 414             $sSQL .= "    ) AS addressimportance, ";
 415         }
 416         $sSQL .= "    (extratags->'place') AS extra_place ";
 417         $sSQL .= " FROM placex";
 418         $sSQL .= " WHERE place_id in ($sPlaceIDs) ";
 419         $sSQL .= "   AND (";
 420         $sSQL .= "            placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
 421         if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) {
 422             $sSQL .= "        OR (extratags->'place') = 'city'";
 423         }
 424         if ($this->aAddressRankList) {
 425             $sSQL .= "        OR placex.rank_address in (".join(',', $this->aAddressRankList).")";
 426         }
 427         $sSQL .= "       ) ";
 428         if ($this->sAllowedTypesSQLList) {
 429             $sSQL .= "AND placex.class in $this->sAllowedTypesSQLList ";
 430         }
 431         $sSQL .= "    AND linked_place_id is null ";
 432         $sSQL .= " GROUP BY ";
 433         $sSQL .= "     osm_type, ";
 434         $sSQL .= "     osm_id, ";
 435         $sSQL .= "     class, ";
 436         $sSQL .= "     type, ";
 437         $sSQL .= "     admin_level, ";
 438         $sSQL .= "     rank_search, ";
 439         $sSQL .= "     rank_address, ";
 440         $sSQL .= "     country_code, ";
 441         $sSQL .= "     importance, ";
 442         if (!$this->bDeDupe) $sSQL .= "place_id,";
 443         $sSQL .= "     langaddress, ";
 444         $sSQL .= "     placename, ";
 445         $sSQL .= "     ref, ";
 446         if ($this->bIncludeExtraTags) $sSQL .= "extratags, ";
 447         if ($this->bIncludeNameDetails) $sSQL .= "name, ";
 448         $sSQL .= "     extratags->'place' ";
 449
 450         // postcode table
 451         $sSQL .= "UNION ";
 452         $sSQL .= "SELECT";
 453         $sSQL .= "  'P' as osm_type,";
 454         $sSQL .= "  (SELECT osm_id from placex p WHERE p.place_id = lp.parent_place_id) as osm_id,";
 455         $sSQL .= "  'place' as class, 'postcode' as type,";
 456         $sSQL .= "  null as admin_level, rank_search, rank_address,";
 457         $sSQL .= "  place_id, parent_place_id, country_code,";
 458         $sSQL .= "  get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) AS langaddress,";
 459         $sSQL .= "  postcode as placename,";
 460         $sSQL .= "  postcode as ref,";
 461         if ($this->bIncludeExtraTags) $sSQL .= "null AS extra,";
 462         if ($this->bIncludeNameDetails) $sSQL .= "null AS names,";
 463         $sSQL .= "  ST_x(st_centroid(geometry)) AS lon, ST_y(st_centroid(geometry)) AS lat,";
 464         $sSQL .= "  (0.75-(rank_search::float/40)) $sImportanceSQLGeom AS importance, ";
 465         if ($oCtx->hasNearPoint()) {
 466             $sSQL .= $oCtx->distanceSQL('geometry')." AS addressimportance,";
 467         } else {
 468             $sSQL .= "  (";
 469             $sSQL .= "     SELECT max(p.importance*(p.rank_address+2))";
 470             $sSQL .= "     FROM ";
 471             $sSQL .= "       place_addressline s, ";
 472             $sSQL .= "       placex p";
 473             $sSQL .= "     WHERE s.place_id = lp.parent_place_id";
 474             $sSQL .= "       AND p.place_id = s.address_place_id ";
 475             $sSQL .= "       AND s.isaddress";
 476             $sSQL .= "       AND p.importance is not null";
 477             $sSQL .= "  ) AS addressimportance, ";
 478         }
 479         $sSQL .= "  null AS extra_place ";
 480         $sSQL .= "FROM location_postcode lp";
 481         $sSQL .= " WHERE place_id in ($sPlaceIDs) ";
 482
 483         if (30 >= $this->iMinAddressRank && 30 <= $this->iMaxAddressRank) {
 484             // only Tiger housenumbers and interpolation lines need to be interpolated, because they are saved as lines
 485             // with start- and endnumber, the common osm housenumbers are usually saved as points
 486             $sHousenumbers = "";
 487             $i = 0;
 488             $length = count($aPlaceIDs);
 489             foreach ($aPlaceIDs as $placeID => $housenumber) {
 490                 $i++;
 491                 $sHousenumbers .= "(".$placeID.", ".$housenumber.")";
 492                 if ($i<$length) $sHousenumbers .= ", ";
 493             }
 494
 495             if (CONST_Use_US_Tiger_Data) {
 496                 // Tiger search only if a housenumber was searched and if it was found (i.e. aPlaceIDs[placeID] = housenumber != -1) (realized through a join)
 497                 $sSQL .= " union";
 498                 $sSQL .= " SELECT ";
 499                 $sSQL .= "     'T' AS osm_type, ";
 500                 $sSQL .= "     (SELECT osm_id from placex p WHERE p.place_id=min(blub.parent_place_id)) as osm_id, ";
 501                 $sSQL .= "     'place' AS class, ";
 502                 $sSQL .= "     'house' AS type, ";
 503                 $sSQL .= "     null AS admin_level, ";
 504                 $sSQL .= "     30 AS rank_search, ";
 505                 $sSQL .= "     30 AS rank_address, ";
 506                 $sSQL .= "     min(place_id) AS place_id, ";
 507                 $sSQL .= "     min(parent_place_id) AS parent_place_id, ";
 508                 $sSQL .= "     'us' AS country_code, ";
 509                 $sSQL .= "     get_address_by_language(place_id, housenumber_for_place, $sLanguagePrefArraySQL) AS langaddress,";
 510                 $sSQL .= "     null AS placename, ";
 511                 $sSQL .= "     null AS ref, ";
 512                 if ($this->bIncludeExtraTags) $sSQL .= "null AS extra,";
 513                 if ($this->bIncludeNameDetails) $sSQL .= "null AS names,";
 514                 $sSQL .= "     avg(st_x(centroid)) AS lon, ";
 515                 $sSQL .= "     avg(st_y(centroid)) AS lat,";
 516                 $sSQL .= "     -1.15".$sImportanceSQL." AS importance, ";
 517                 if ($oCtx->hasNearPoint()) {
 518                     $sSQL .= $oCtx->distanceSQL('ST_Collect(centroid)')." AS addressimportance,";
 519                 } else {
 520                     $sSQL .= "     (";
 521                     $sSQL .= "        SELECT max(p.importance*(p.rank_address+2))";
 522                     $sSQL .= "        FROM ";
 523                     $sSQL .= "          place_addressline s, ";
 524                     $sSQL .= "          placex p";
 525                     $sSQL .= "        WHERE s.place_id = min(blub.parent_place_id)";
 526                     $sSQL .= "          AND p.place_id = s.address_place_id ";
 527                     $sSQL .= "          AND s.isaddress";
 528                     $sSQL .= "          AND p.importance is not null";
 529                     $sSQL .= "     ) AS addressimportance, ";
 530                 }
 531                 $sSQL .= "     null AS extra_place ";
 532                 $sSQL .= " FROM (";
 533                 $sSQL .= "     SELECT place_id, ";    // interpolate the Tiger housenumbers here
 534                 $sSQL .= "         ST_LineInterpolatePoint(linegeo, (housenumber_for_place-startnumber::float)/(endnumber-startnumber)::float) AS centroid, ";
 535                 $sSQL .= "         parent_place_id, ";
 536                 $sSQL .= "         housenumber_for_place";
 537                 $sSQL .= "     FROM (";
 538                 $sSQL .= "            location_property_tiger ";
 539                 $sSQL .= "            JOIN (values ".$sHousenumbers.") AS housenumbers(place_id, housenumber_for_place) USING(place_id)) ";
 540                 $sSQL .= "     WHERE ";
 541                 $sSQL .= "         housenumber_for_place>=0";
 542                 $sSQL .= "         AND 30 between $this->iMinAddressRank AND $this->iMaxAddressRank";
 543                 $sSQL .= " ) AS blub"; //postgres wants an alias here
 544                 $sSQL .= " GROUP BY";
 545                 $sSQL .= "      place_id, ";
 546                 $sSQL .= "      housenumber_for_place"; //is this group by really needed?, place_id + housenumber (in combination) are unique
 547                 if (!$this->bDeDupe) $sSQL .= ", place_id ";
 548             }
 549             // osmline
 550             // interpolation line search only if a housenumber was searched and if it was found (i.e. aPlaceIDs[placeID] = housenumber != -1) (realized through a join)
 551             $sSQL .= " UNION ";
 552             $sSQL .= "SELECT ";
 553             $sSQL .= "  'W' AS osm_type, ";
 554             $sSQL .= "  osm_id, ";
 555             $sSQL .= "  'place' AS class, ";
 556             $sSQL .= "  'house' AS type, ";
 557             $sSQL .= "  null AS admin_level, ";
 558             $sSQL .= "  30 AS rank_search, ";
 559             $sSQL .= "  30 AS rank_address, ";
 560             $sSQL .= "  min(place_id) as place_id, ";
 561             $sSQL .= "  min(parent_place_id) AS parent_place_id, ";
 562             $sSQL .= "  country_code, ";
 563             $sSQL .= "  get_address_by_language(place_id, housenumber_for_place, $sLanguagePrefArraySQL) AS langaddress, ";
 564             $sSQL .= "  null AS placename, ";
 565             $sSQL .= "  null AS ref, ";
 566             if ($this->bIncludeExtraTags) $sSQL .= "null AS extra, ";
 567             if ($this->bIncludeNameDetails) $sSQL .= "null AS names, ";
 568             $sSQL .= "  AVG(st_x(centroid)) AS lon, ";
 569             $sSQL .= "  AVG(st_y(centroid)) AS lat, ";
 570             $sSQL .= "  -0.1".$sImportanceSQL." AS importance, ";  // slightly smaller than the importance for normal houses with rank 30, which is 0
 571             if ($oCtx->hasNearPoint()) {
 572                 $sSQL .= $oCtx->distanceSQL('ST_Collect(centroid)')." AS addressimportance,";
 573             } else {
 574                 $sSQL .= "  (";
 575                 $sSQL .= "     SELECT ";
 576                 $sSQL .= "       MAX(p.importance*(p.rank_address+2)) ";
 577                 $sSQL .= "     FROM";
 578                 $sSQL .= "       place_addressline s, ";
 579                 $sSQL .= "       placex p";
 580                 $sSQL .= "     WHERE s.place_id = min(blub.parent_place_id) ";
 581                 $sSQL .= "       AND p.place_id = s.address_place_id ";
 582                 $sSQL .= "       AND s.isaddress ";
 583                 $sSQL .= "       AND p.importance is not null";
 584                 $sSQL .= "  ) AS addressimportance,";
 585             }
 586             $sSQL .= "  null AS extra_place ";
 587             $sSQL .= "  FROM (";
 588             $sSQL .= "     SELECT ";
 589             $sSQL .= "         osm_id, ";
 590             $sSQL .= "         place_id, ";
 591             $sSQL .= "         country_code, ";
 592             $sSQL .= "         CASE ";             // interpolate the housenumbers here
 593             $sSQL .= "           WHEN startnumber != endnumber ";
 594             $sSQL .= "           THEN ST_LineInterpolatePoint(linegeo, (housenumber_for_place-startnumber::float)/(endnumber-startnumber)::float) ";
 595             $sSQL .= "           ELSE ST_LineInterpolatePoint(linegeo, 0.5) ";
 596             $sSQL .= "         END as centroid, ";
 597             $sSQL .= "         parent_place_id, ";
 598             $sSQL .= "         housenumber_for_place ";
 599             $sSQL .= "     FROM (";
 600             $sSQL .= "            location_property_osmline ";
 601             $sSQL .= "            JOIN (values ".$sHousenumbers.") AS housenumbers(place_id, housenumber_for_place) USING(place_id)";
 602             $sSQL .= "          ) ";
 603             $sSQL .= "     WHERE housenumber_for_place>=0 ";
 604             $sSQL .= "       AND 30 between $this->iMinAddressRank AND $this->iMaxAddressRank";
 605             $sSQL .= "  ) as blub"; //postgres wants an alias here
 606             $sSQL .= "  GROUP BY ";
 607             $sSQL .= "    osm_id, ";
 608             $sSQL .= "    place_id, ";
 609             $sSQL .= "    housenumber_for_place, ";
 610             $sSQL .= "    country_code "; //is this group by really needed?, place_id + housenumber (in combination) are unique
 611             if (!$this->bDeDupe) $sSQL .= ", place_id ";
 612
 613             if (CONST_Use_Aux_Location_data) {
 614                 $sSQL .= " UNION ";
 615                 $sSQL .= "  SELECT ";
 616                 $sSQL .= "     'L' AS osm_type, ";
 617                 $sSQL .= "     place_id AS osm_id, ";
 618                 $sSQL .= "     'place' AS class,";
 619                 $sSQL .= "     'house' AS type, ";
 620                 $sSQL .= "     null AS admin_level, ";
 621                 $sSQL .= "     0 AS rank_search,";
 622                 $sSQL .= "     0 AS rank_address, ";
 623                 $sSQL .= "     min(place_id) AS place_id,";
 624                 $sSQL .= "     min(parent_place_id) AS parent_place_id, ";
 625                 $sSQL .= "     'us' AS country_code, ";
 626                 $sSQL .= "     get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) AS langaddress, ";
 627                 $sSQL .= "     null AS placename, ";
 628                 $sSQL .= "     null AS ref, ";
 629                 if ($this->bIncludeExtraTags) $sSQL .= "null AS extra, ";
 630                 if ($this->bIncludeNameDetails) $sSQL .= "null AS names, ";
 631                 $sSQL .= "     avg(ST_X(centroid)) AS lon, ";
 632                 $sSQL .= "     avg(ST_Y(centroid)) AS lat, ";
 633                 $sSQL .= "     -1.10".$sImportanceSQL." AS importance, ";
 634                 if ($oCtx->hasNearPoint()) {
 635                     $sSQL .= $oCtx->distanceSQL('ST_Collect(centroid)')." AS addressimportance,";
 636                 } else {
 637                     $sSQL .= "     ( ";
 638                     $sSQL .= "       SELECT max(p.importance*(p.rank_address+2))";
 639                     $sSQL .= "       FROM ";
 640                     $sSQL .= "          place_addressline s, ";
 641                     $sSQL .= "          placex p";
 642                     $sSQL .= "       WHERE s.place_id = min(location_property_aux.parent_place_id)";
 643                     $sSQL .= "         AND p.place_id = s.address_place_id ";
 644                     $sSQL .= "         AND s.isaddress";
 645                     $sSQL .= "         AND p.importance is not null";
 646                     $sSQL .= "     ) AS addressimportance, ";
 647                 }
 648                 $sSQL .= "     null AS extra_place ";
 649                 $sSQL .= "  FROM location_property_aux ";
 650                 $sSQL .= "  WHERE place_id in ($sPlaceIDs) ";
 651                 $sSQL .= "    AND 30 between $this->iMinAddressRank and $this->iMaxAddressRank ";
 652                 $sSQL .= "  GROUP BY ";
 653                 $sSQL .= "     place_id, ";
 654                 if (!$this->bDeDupe) $sSQL .= "place_id, ";
 655                 $sSQL .= "     get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) ";
 656             }
 657         }
 658
 659         $sSQL .= " order by importance desc";
 660         if (CONST_Debug) {
 661             echo "<hr>";
 662             var_dump($sSQL);
 663         }
 664         $aSearchResults = chksql(
 665             $this->oDB->getAll($sSQL),
 666             "Could not get details for place."
 667         );
 668
 669         return $aSearchResults;
 670     }
 671
 672     public function getGroupedSearches($aSearches, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bIsStructured, $sNormQuery)
 673     {
 674         /*
 675              Calculate all searches using aValidTokens i.e.
 676              'Wodsworth Road, Sheffield' =>
 677
 678              Phrase Wordset
 679              0      0       (wodsworth road)
 680              0      1       (wodsworth)(road)
 681              1      0       (sheffield)
 682
 683              Score how good the search is so they can be ordered
 684          */
 685         $iGlobalRank = 0;
 686
 687         foreach ($aPhrases as $iPhrase => $oPhrase) {
 688             $aNewPhraseSearches = array();
 689             $sPhraseType = $bIsStructured ? $oPhrase->getPhraseType() : '';
 690
 691             foreach ($oPhrase->getWordSets() as $iWordSet => $aWordset) {
 692                 // Too many permutations - too expensive
 693                 if ($iWordSet > 120) break;
 694
 695                 $aWordsetSearches = $aSearches;
 696
 697                 // Add all words from this wordset
 698                 foreach ($aWordset as $iToken => $sToken) {
 699                     //echo "<br><b>$sToken</b>";
 700                     $aNewWordsetSearches = array();
 701
 702                     foreach ($aWordsetSearches as $oCurrentSearch) {
 703                         //echo "<i>";
 704                         //var_dump($oCurrentSearch);
 705                         //echo "</i>";
 706
 707                         // If the token is valid
 708                         if (isset($aValidTokens[' '.$sToken])) {
 709                             foreach ($aValidTokens[' '.$sToken] as $aSearchTerm) {
 710                                 // Recheck if the original word shows up in the query.
 711                                 $bWordInQuery = false;
 712                                 if (isset($aSearchTerm['word']) && $aSearchTerm['word']) {
 713                                     $bWordInQuery = strpos(
 714                                         $sNormQuery,
 715                                         $this->normTerm($aSearchTerm['word'])
 716                                     ) !== false;
 717                                 }
 718                                 $aNewSearches = $oCurrentSearch->extendWithFullTerm(
 719                                     $aSearchTerm,
 720                                     $bWordInQuery,
 721                                     isset($aValidTokens[$sToken])
 722                                       && strpos($sToken, ' ') === false,
 723                                     $sPhraseType,
 724                                     $iToken == 0 && $iPhrase == 0,
 725                                     $iPhrase == 0,
 726                                     $iToken + 1 == sizeof($aWordset)
 727                                       && $iPhrase + 1 == sizeof($aPhrases),
 728                                     $iGlobalRank
 729                                 );
 730
 731                                 foreach ($aNewSearches as $oSearch) {
 732                                     if ($oSearch->getRank() < $this->iMaxRank) {
 733                                         $aNewWordsetSearches[] = $oSearch;
 734                                     }
 735                                 }
 736                             }
 737                         }
 738                         // Look for partial matches.
 739                         // Note that there is no point in adding country terms here
 740                         // because country is omitted in the address.
 741                         if (isset($aValidTokens[$sToken]) && $sPhraseType != 'country') {
 742                             // Allow searching for a word - but at extra cost
 743                             foreach ($aValidTokens[$sToken] as $aSearchTerm) {
 744                                 $aNewSearches = $oCurrentSearch->extendWithPartialTerm(
 745                                     $aSearchTerm,
 746                                     $bIsStructured,
 747                                     $iPhrase,
 748                                     $aWordFrequencyScores,
 749                                     isset($aValidTokens[' '.$sToken]) ? $aValidTokens[' '.$sToken] : array()
 750                                 );
 751
 752                                 foreach ($aNewSearches as $oSearch) {
 753                                     if ($oSearch->getRank() < $this->iMaxRank) {
 754                                         $aNewWordsetSearches[] = $oSearch;
 755                                     }
 756                                 }
 757                             }
 758                         }
 759                     }
 760                     // Sort and cut
 761                     usort($aNewWordsetSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 762                     $aWordsetSearches = array_slice($aNewWordsetSearches, 0, 50);
 763                 }
 764                 //var_Dump('<hr>',sizeof($aWordsetSearches)); exit;
 765
 766                 $aNewPhraseSearches = array_merge($aNewPhraseSearches, $aNewWordsetSearches);
 767                 usort($aNewPhraseSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 768
 769                 $aSearchHash = array();
 770                 foreach ($aNewPhraseSearches as $iSearch => $aSearch) {
 771                     $sHash = serialize($aSearch);
 772                     if (isset($aSearchHash[$sHash])) unset($aNewPhraseSearches[$iSearch]);
 773                     else $aSearchHash[$sHash] = 1;
 774                 }
 775
 776                 $aNewPhraseSearches = array_slice($aNewPhraseSearches, 0, 50);
 777             }
 778
 779             // Re-group the searches by their score, junk anything over 20 as just not worth trying
 780             $aGroupedSearches = array();
 781             foreach ($aNewPhraseSearches as $aSearch) {
 782                 $iRank = $aSearch->getRank();
 783                 if ($iRank < $this->iMaxRank) {
 784                     if (!isset($aGroupedSearches[$iRank])) {
 785                         $aGroupedSearches[$iRank] = array();
 786                     }
 787                     $aGroupedSearches[$iRank][] = $aSearch;
 788                 }
 789             }
 790             ksort($aGroupedSearches);
 791
 792             $iSearchCount = 0;
 793             $aSearches = array();
 794             foreach ($aGroupedSearches as $iScore => $aNewSearches) {
 795                 $iSearchCount += sizeof($aNewSearches);
 796                 $aSearches = array_merge($aSearches, $aNewSearches);
 797                 if ($iSearchCount > 50) break;
 798             }
 799
 800             //if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens);
 801         }
 802
 803         // Revisit searches, drop bad searches and give penalty to unlikely combinations.
 804         $aGroupedSearches = array();
 805         foreach ($aSearches as $oSearch) {
 806             if (!$oSearch->isValidSearch($this->aCountryCodes)) {
 807                 continue;
 808             }
 809
 810             $iRank = $oSearch->addToRank($iGlobalRank);
 811             if (!isset($aGroupedSearches[$iRank])) {
 812                 $aGroupedSearches[$iRank] = array();
 813             }
 814             $aGroupedSearches[$iRank][] = $oSearch;
 815         }
 816         ksort($aGroupedSearches);
 817
 818         return $aGroupedSearches;
 819     }
 820
 821     /* Perform the actual query lookup.
 822
 823         Returns an ordered list of results, each with the following fields:
 824             osm_type: type of corresponding OSM object
 825                         N - node
 826                         W - way
 827                         R - relation
 828                         P - postcode (internally computed)
 829             osm_id: id of corresponding OSM object
 830             class: general object class (corresponds to tag key of primary OSM tag)
 831             type: subclass of object (corresponds to tag value of primary OSM tag)
 832             admin_level: see http://wiki.openstreetmap.org/wiki/Admin_level
 833             rank_search: rank in search hierarchy
 834                         (see also http://wiki.openstreetmap.org/wiki/Nominatim/Development_overview#Country_to_street_level)
 835             rank_address: rank in address hierarchy (determines orer in address)
 836             place_id: internal key (may differ between different instances)
 837             country_code: ISO country code
 838             langaddress: localized full address
 839             placename: localized name of object
 840             ref: content of ref tag (if available)
 841             lon: longitude
 842             lat: latitude
 843             importance: importance of place based on Wikipedia link count
 844             addressimportance: cumulated importance of address elements
 845             extra_place: type of place (for admin boundaries, if there is a place tag)
 846             aBoundingBox: bounding Box
 847             label: short description of the object class/type (English only)
 848             name: full name (currently the same as langaddress)
 849             foundorder: secondary ordering for places with same importance
 850     */
 851
 852
 853     public function lookup()
 854     {
 855         if (!$this->sQuery && !$this->aStructuredQuery) return array();
 856
 857         $oCtx = new SearchContext();
 858
 859         if ($this->aRoutePoints) {
 860             $oCtx->setViewboxFromRoute(
 861                 $this->oDB,
 862                 $this->aRoutePoints,
 863                 $this->aRouteWidth,
 864                 $this->bBoundedSearch
 865             );
 866         } elseif ($this->aViewBox) {
 867             $oCtx->setViewboxFromBox($this->aViewBox, $this->bBoundedSearch);
 868         }
 869         if ($this->aExcludePlaceIDs) {
 870             $oCtx->setExcludeList($this->aExcludePlaceIDs);
 871         }
 872         if ($this->aCountryCodes) {
 873             $oCtx->setCountryList($this->aCountryCodes);
 874         }
 875
 876         $sNormQuery = $this->normTerm($this->sQuery);
 877         $sLanguagePrefArraySQL = getArraySQL(
 878             array_map("getDBQuoted", $this->aLangPrefOrder)
 879         );
 880
 881         $sQuery = $this->sQuery;
 882         if (!preg_match('//u', $sQuery)) {
 883             userError("Query string is not UTF-8 encoded.");
 884         }
 885
 886         // Conflicts between US state abreviations and various words for 'the' in different languages
 887         if (isset($this->aLangPrefOrder['name:en'])) {
 888             $sQuery = preg_replace('/(^|,)\s*il\s*(,|$)/', '\1illinois\2', $sQuery);
 889             $sQuery = preg_replace('/(^|,)\s*al\s*(,|$)/', '\1alabama\2', $sQuery);
 890             $sQuery = preg_replace('/(^|,)\s*la\s*(,|$)/', '\1louisiana\2', $sQuery);
 891         }
 892
 893         // Do we have anything that looks like a lat/lon pair?
 894         $sQuery = $oCtx->setNearPointFromQuery($sQuery);
 895
 896         $aSearchResults = array();
 897         if ($sQuery || $this->aStructuredQuery) {
 898             // Start with a single blank search
 899             $aSearches = array(new SearchDescription($oCtx));
 900
 901             if ($sQuery) {
 902                 $sQuery = $aSearches[0]->extractKeyValuePairs($sQuery);
 903             }
 904
 905             $sSpecialTerm = '';
 906             if ($sQuery) {
 907                 preg_match_all(
 908                     '/\\[([\\w ]*)\\]/u',
 909                     $sQuery,
 910                     $aSpecialTermsRaw,
 911                     PREG_SET_ORDER
 912                 );
 913                 foreach ($aSpecialTermsRaw as $aSpecialTerm) {
 914                     $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
 915                     if (!$sSpecialTerm) {
 916                         $sSpecialTerm = $aSpecialTerm[1];
 917                     }
 918                 }
 919             }
 920             if (!$sSpecialTerm && $this->aStructuredQuery
 921                 && isset($this->aStructuredQuery['amenity'])) {
 922                 $sSpecialTerm = $this->aStructuredQuery['amenity'];
 923                 unset($this->aStructuredQuery['amenity']);
 924             }
 925
 926             if ($sSpecialTerm && !$aSearches[0]->hasOperator()) {
 927                 $sSpecialTerm = pg_escape_string($sSpecialTerm);
 928                 $sToken = chksql(
 929                     $this->oDB->getOne("SELECT make_standard_name('$sSpecialTerm')"),
 930                     "Cannot decode query. Wrong encoding?"
 931                 );
 932                 $sSQL = 'SELECT class, type FROM word ';
 933                 $sSQL .= '   WHERE word_token in (\' '.$sToken.'\')';
 934                 $sSQL .= '   AND class is not null AND class not in (\'place\')';
 935                 if (CONST_Debug) var_Dump($sSQL);
 936                 $aSearchWords = chksql($this->oDB->getAll($sSQL));
 937                 $aNewSearches = array();
 938                 foreach ($aSearches as $oSearch) {
 939                     foreach ($aSearchWords as $aSearchTerm) {
 940                         $oNewSearch = clone $oSearch;
 941                         $oNewSearch->setPoiSearch(
 942                             Operator::TYPE,
 943                             $aSearchTerm['class'],
 944                             $aSearchTerm['type']
 945                         );
 946                         $aNewSearches[] = $oNewSearch;
 947                     }
 948                 }
 949                 $aSearches = $aNewSearches;
 950             }
 951
 952             // Split query into phrases
 953             // Commas are used to reduce the search space by indicating where phrases split
 954             if ($this->aStructuredQuery) {
 955                 $aInPhrases = $this->aStructuredQuery;
 956                 $bStructuredPhrases = true;
 957             } else {
 958                 $aInPhrases = explode(',', $sQuery);
 959                 $bStructuredPhrases = false;
 960             }
 961
 962             // Convert each phrase to standard form
 963             // Create a list of standard words
 964             // Get all 'sets' of words
 965             // Generate a complete list of all
 966             $aTokens = array();
 967             $aPhrases = array();
 968             foreach ($aInPhrases as $iPhrase => $sPhrase) {
 969                 $sPhrase = chksql(
 970                     $this->oDB->getOne('SELECT make_standard_name('.getDBQuoted($sPhrase).')'),
 971                     "Cannot normalize query string (is it a UTF-8 string?)"
 972                 );
 973                 if (trim($sPhrase)) {
 974                     $oPhrase = new Phrase($sPhrase, is_string($iPhrase) ? $iPhrase : '');
 975                     $oPhrase->addTokens($aTokens);
 976                     $aPhrases[] = $oPhrase;
 977                 }
 978             }
 979
 980             if (sizeof($aTokens)) {
 981                 // Check which tokens we have, get the ID numbers
 982                 $sSQL = 'SELECT word_id, word_token, word, class, type, country_code, operator, search_name_count';
 983                 $sSQL .= ' FROM word ';
 984                 $sSQL .= ' WHERE word_token in ('.join(',', array_map("getDBQuoted", $aTokens)).')';
 985
 986                 if (CONST_Debug) var_Dump($sSQL);
 987
 988                 $aValidTokens = array();
 989                 $aDatabaseWords = chksql(
 990                     $this->oDB->getAll($sSQL),
 991                     "Could not get word tokens."
 992                 );
 993                 $aPossibleMainWordIDs = array();
 994                 $aWordFrequencyScores = array();
 995                 foreach ($aDatabaseWords as $aToken) {
 996                     // Very special case - require 2 letter country param to match the country code found
 997                     if ($bStructuredPhrases && $aToken['country_code'] && !empty($this->aStructuredQuery['country'])
 998                         && strlen($this->aStructuredQuery['country']) == 2 && strtolower($this->aStructuredQuery['country']) != $aToken['country_code']
 999                     ) {
1000                         continue;
1001                     }
1002
1003                     if (isset($aValidTokens[$aToken['word_token']])) {
1004                         $aValidTokens[$aToken['word_token']][] = $aToken;
1005                     } else {
1006                         $aValidTokens[$aToken['word_token']] = array($aToken);
1007                     }
1008                     if (!$aToken['class'] && !$aToken['country_code']) $aPossibleMainWordIDs[$aToken['word_id']] = 1;
1009                     $aWordFrequencyScores[$aToken['word_id']] = $aToken['search_name_count'] + 1;
1010                 }
1011                 if (CONST_Debug) var_Dump($aPhrases, $aValidTokens);
1012
1013                 // US ZIP+4 codes - if there is no token, merge in the 5-digit ZIP code
1014                 foreach ($aTokens as $sToken) {
1015                     if (!isset($aValidTokens[$sToken]) && preg_match('/^([0-9]{5}) [0-9]{4}$/', $sToken, $aData)) {
1016                         if (isset($aValidTokens[$aData[1]])) {
1017                             foreach ($aValidTokens[$aData[1]] as $aToken) {
1018                                 if (!$aToken['class']) {
1019                                     if (isset($aValidTokens[$sToken])) {
1020                                         $aValidTokens[$sToken][] = $aToken;
1021                                     } else {
1022                                         $aValidTokens[$sToken] = array($aToken);
1023                                     }
1024                                 }
1025                             }
1026                         }
1027                     }
1028                 }
1029
1030                 foreach ($aTokens as $sToken) {
1031                     // Unknown single word token with a number - assume it is a house number
1032                     if (!isset($aValidTokens[' '.$sToken]) && strpos($sToken, ' ') === false && preg_match('/^[0-9]+$/', $sToken)) {
1033                         $aValidTokens[' '.$sToken] = array(array('class' => 'place', 'type' => 'house', 'word_token' => ' '.$sToken));
1034                     }
1035                 }
1036
1037                 // Any words that have failed completely?
1038                 // TODO: suggestions
1039
1040                 $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases, $sNormQuery);
1041
1042                 if ($this->bReverseInPlan) {
1043                     // Reverse phrase array and also reverse the order of the wordsets in
1044                     // the first and final phrase. Don't bother about phrases in the middle
1045                     // because order in the address doesn't matter.
1046                     $aPhrases = array_reverse($aPhrases);
1047                     $aPhrases[0]->invertWordSets();
1048                     if (sizeof($aPhrases) > 1) {
1049                         $aPhrases[sizeof($aPhrases)-1]->invertWordSets();
1050                     }
1051                     $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $aValidTokens, $aWordFrequencyScores, false, $sNormQuery);
1052
1053                     foreach ($aGroupedSearches as $aSearches) {
1054                         foreach ($aSearches as $aSearch) {
1055                             if (!isset($aReverseGroupedSearches[$aSearch->getRank()])) {
1056                                 $aReverseGroupedSearches[$aSearch->getRank()] = array();
1057                             }
1058                             $aReverseGroupedSearches[$aSearch->getRank()][] = $aSearch;
1059                         }
1060                     }
1061
1062                     $aGroupedSearches = $aReverseGroupedSearches;
1063                     ksort($aGroupedSearches);
1064                 }
1065             } else {
1066                 // Re-group the searches by their score, junk anything over 20 as just not worth trying
1067                 $aGroupedSearches = array();
1068                 foreach ($aSearches as $aSearch) {
1069                     if ($aSearch->getRank() < $this->iMaxRank) {
1070                         if (!isset($aGroupedSearches[$aSearch->getRank()])) $aGroupedSearches[$aSearch->getRank()] = array();
1071                         $aGroupedSearches[$aSearch->getRank()][] = $aSearch;
1072                     }
1073                 }
1074                 ksort($aGroupedSearches);
1075             }
1076
1077             // Filter out duplicate searches
1078             $aSearchHash = array();
1079             foreach ($aGroupedSearches as $iGroup => $aSearches) {
1080                 foreach ($aSearches as $iSearch => $aSearch) {
1081                     $sHash = serialize($aSearch);
1082                     if (isset($aSearchHash[$sHash])) {
1083                         unset($aGroupedSearches[$iGroup][$iSearch]);
1084                         if (sizeof($aGroupedSearches[$iGroup]) == 0) unset($aGroupedSearches[$iGroup]);
1085                     } else {
1086                         $aSearchHash[$sHash] = 1;
1087                     }
1088                 }
1089             }
1090
1091             if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens);
1092
1093             // Start the search process
1094             // array with: placeid => -1 | tiger-housenumber
1095             $aResultPlaceIDs = array();
1096             $iGroupLoop = 0;
1097             $iQueryLoop = 0;
1098             foreach ($aGroupedSearches as $iGroupedRank => $aSearches) {
1099                 $iGroupLoop++;
1100                 foreach ($aSearches as $oSearch) {
1101                     $iQueryLoop++;
1102
1103                     if (CONST_Debug) {
1104                         echo "<hr><b>Search Loop, group $iGroupLoop, loop $iQueryLoop</b>";
1105                         _debugDumpGroupedSearches(array($iGroupedRank => array($oSearch)), $aValidTokens);
1106                     }
1107
1108                     $aRes = $oSearch->query(
1109                         $this->oDB,
1110                         $aWordFrequencyScores,
1111                         $this->exactMatchCache,
1112                         $this->iMinAddressRank,
1113                         $this->iMaxAddressRank,
1114                         $this->iLimit
1115                     );
1116
1117                     foreach ($aRes['IDs'] as $iPlaceID) {
1118                         // array for placeID => -1 | Tiger housenumber
1119                         $aResultPlaceIDs[$iPlaceID] = $aRes['houseNumber'];
1120                     }
1121                     if ($iQueryLoop > 20) break;
1122                 }
1123
1124                 if (sizeof($aResultPlaceIDs) && ($this->iMinAddressRank != 0 || $this->iMaxAddressRank != 30)) {
1125                     // Need to verify passes rank limits before dropping out of the loop (yuk!)
1126                     // reduces the number of place ids, like a filter
1127                     // rank_address is 30 for interpolated housenumbers
1128                     $sWherePlaceId = 'WHERE place_id in (';
1129                     $sWherePlaceId .= join(',', array_keys($aResultPlaceIDs)).') ';
1130
1131                     $sSQL = "SELECT place_id ";
1132                     $sSQL .= "FROM placex ".$sWherePlaceId;
1133                     $sSQL .= "  AND (";
1134                     $sSQL .= "         placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
1135                     if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) {
1136                         $sSQL .= "     OR (extratags->'place') = 'city'";
1137                     }
1138                     if ($this->aAddressRankList) {
1139                         $sSQL .= "     OR placex.rank_address in (".join(',', $this->aAddressRankList).")";
1140                     }
1141                     $sSQL .= "  ) UNION ";
1142                     $sSQL .= " SELECT place_id FROM location_postcode lp ".$sWherePlaceId;
1143                     $sSQL .= "  AND (lp.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
1144                     if ($this->aAddressRankList) {
1145                         $sSQL .= "     OR lp.rank_address in (".join(',', $this->aAddressRankList).")";
1146                     }
1147                     $sSQL .= ") ";
1148                     if (CONST_Use_US_Tiger_Data && $this->iMaxAddressRank == 30) {
1149                         $sSQL .= "UNION ";
1150                         $sSQL .= "  SELECT place_id ";
1151                         $sSQL .= "  FROM location_property_tiger ".$sWherePlaceId;
1152                     }
1153                     if ($this->iMaxAddressRank == 30) {
1154                         $sSQL .= "UNION ";
1155                         $sSQL .= "  SELECT place_id ";
1156                         $sSQL .= "  FROM location_property_osmline ".$sWherePlaceId;
1157                     }
1158                     if (CONST_Debug) var_dump($sSQL);
1159                     $aFilteredPlaceIDs = chksql($this->oDB->getCol($sSQL));
1160                     $tempIDs = array();
1161                     foreach ($aFilteredPlaceIDs as $placeID) {
1162                         $tempIDs[$placeID] = $aResultPlaceIDs[$placeID];  //assign housenumber to placeID
1163                     }
1164                     $aResultPlaceIDs = $tempIDs;
1165                 }
1166
1167                 if (sizeof($aResultPlaceIDs)) break;
1168                 if ($iGroupLoop > 4) break;
1169                 if ($iQueryLoop > 30) break;
1170             }
1171
1172             // Did we find anything?
1173             if (sizeof($aResultPlaceIDs)) {
1174                 $aSearchResults = $this->getDetails($aResultPlaceIDs, $oCtx);
1175             }
1176         } else {
1177             // Just interpret as a reverse geocode
1178             $oReverse = new ReverseGeocode($this->oDB);
1179             $oReverse->setZoom(18);
1180
1181             $aLookup = $oReverse->lookupPoint($oCtx->sqlNear, false);
1182
1183             if (CONST_Debug) var_dump("Reverse search", $aLookup);
1184
1185             if ($aLookup['place_id']) {
1186                 $aSearchResults = $this->getDetails(array($aLookup['place_id'] => -1), $oCtx);
1187                 $aResultPlaceIDs[$aLookup['place_id']] = -1;
1188             } else {
1189                 $aSearchResults = array();
1190             }
1191         }
1192
1193         // No results? Done
1194         if (!sizeof($aSearchResults)) {
1195             if ($this->bFallback) {
1196                 if ($this->fallbackStructuredQuery()) {
1197                     return $this->lookup();
1198                 }
1199             }
1200
1201             return array();
1202         }
1203
1204         $aClassType = getClassTypesWithImportance();
1205         $aRecheckWords = preg_split('/\b[\s,\\-]*/u', $sQuery);
1206         foreach ($aRecheckWords as $i => $sWord) {
1207             if (!preg_match('/[\pL\pN]/', $sWord)) unset($aRecheckWords[$i]);
1208         }
1209
1210         if (CONST_Debug) {
1211             echo '<i>Recheck words:<\i>';
1212             var_dump($aRecheckWords);
1213         }
1214
1215         $oPlaceLookup = new PlaceLookup($this->oDB);
1216         $oPlaceLookup->setIncludePolygonAsPoints($this->bIncludePolygonAsPoints);
1217         $oPlaceLookup->setIncludePolygonAsText($this->bIncludePolygonAsText);
1218         $oPlaceLookup->setIncludePolygonAsGeoJSON($this->bIncludePolygonAsGeoJSON);
1219         $oPlaceLookup->setIncludePolygonAsKML($this->bIncludePolygonAsKML);
1220         $oPlaceLookup->setIncludePolygonAsSVG($this->bIncludePolygonAsSVG);
1221         $oPlaceLookup->setPolygonSimplificationThreshold($this->fPolygonSimplificationThreshold);
1222
1223         foreach ($aSearchResults as $iResNum => $aResult) {
1224             // Default
1225             $fDiameter = getResultDiameter($aResult);
1226
1227             $aOutlineResult = $oPlaceLookup->getOutlines($aResult['place_id'], $aResult['lon'], $aResult['lat'], $fDiameter/2);
1228             if ($aOutlineResult) {
1229                 $aResult = array_merge($aResult, $aOutlineResult);
1230             }
1231
1232             if ($aResult['extra_place'] == 'city') {
1233                 $aResult['class'] = 'place';
1234                 $aResult['type'] = 'city';
1235                 $aResult['rank_search'] = 16;
1236             }
1237
1238             // Is there an icon set for this type of result?
1239             if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['icon'])
1240                 && $aClassType[$aResult['class'].':'.$aResult['type']]['icon']
1241             ) {
1242                 $aResult['icon'] = CONST_Website_BaseURL.'images/mapicons/'.$aClassType[$aResult['class'].':'.$aResult['type']]['icon'].'.p.20.png';
1243             }
1244
1245             if (isset($aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label'])
1246                 && $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label']
1247             ) {
1248                 $aResult['label'] = $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label'];
1249             } elseif (isset($aClassType[$aResult['class'].':'.$aResult['type']]['label'])
1250                 && $aClassType[$aResult['class'].':'.$aResult['type']]['label']
1251             ) {
1252                 $aResult['label'] = $aClassType[$aResult['class'].':'.$aResult['type']]['label'];
1253             }
1254             // if tag '&addressdetails=1' is set in query
1255             if ($this->bIncludeAddressDetails) {
1256                 // getAddressDetails() is defined in lib.php and uses the SQL function get_addressdata in functions.sql
1257                 $aResult['address'] = getAddressDetails($this->oDB, $sLanguagePrefArraySQL, $aResult['place_id'], $aResult['country_code'], $aResultPlaceIDs[$aResult['place_id']]);
1258                 if ($aResult['extra_place'] == 'city' && !isset($aResult['address']['city'])) {
1259                     $aResult['address'] = array_merge(array('city' => array_values($aResult['address'])[0]), $aResult['address']);
1260                 }
1261             }
1262
1263             if ($this->bIncludeExtraTags) {
1264                 if ($aResult['extra']) {
1265                     $aResult['sExtraTags'] = json_decode($aResult['extra']);
1266                 } else {
1267                     $aResult['sExtraTags'] = (object) array();
1268                 }
1269             }
1270
1271             if ($this->bIncludeNameDetails) {
1272                 if ($aResult['names']) {
1273                     $aResult['sNameDetails'] = json_decode($aResult['names']);
1274                 } else {
1275                     $aResult['sNameDetails'] = (object) array();
1276                 }
1277             }
1278
1279             $aResult['name'] = $aResult['langaddress'];
1280
1281             if ($oCtx->hasNearPoint())
1282             {
1283                 $aResult['importance'] = 0.001;
1284                 $aResult['foundorder'] = $aResult['addressimportance'];
1285             } else {
1286                 // Adjust importance for the number of exact string matches in the result
1287                 $aResult['importance'] = max(0.001, $aResult['importance']);
1288                 $iCountWords = 0;
1289                 $sAddress = $aResult['langaddress'];
1290                 foreach ($aRecheckWords as $i => $sWord) {
1291                     if (stripos($sAddress, $sWord)!==false) {
1292                         $iCountWords++;
1293                         if (preg_match("/(^|,)\s*".preg_quote($sWord, '/')."\s*(,|$)/", $sAddress)) $iCountWords += 0.1;
1294                     }
1295                 }
1296
1297                 $aResult['importance'] = $aResult['importance'] + ($iCountWords*0.1); // 0.1 is a completely arbitrary number but something in the range 0.1 to 0.5 would seem right
1298
1299                 // secondary ordering (for results with same importance (the smaller the better):
1300                 // - approximate importance of address parts
1301                 $aResult['foundorder'] = -$aResult['addressimportance']/10;
1302                 // - number of exact matches from the query
1303                 if (isset($this->exactMatchCache[$aResult['place_id']])) {
1304                     $aResult['foundorder'] -= $this->exactMatchCache[$aResult['place_id']];
1305                 } elseif (isset($this->exactMatchCache[$aResult['parent_place_id']])) {
1306                     $aResult['foundorder'] -= $this->exactMatchCache[$aResult['parent_place_id']];
1307                 }
1308                 // - importance of the class/type
1309                 if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['importance'])
1310                     && $aClassType[$aResult['class'].':'.$aResult['type']]['importance']
1311                 ) {
1312                     $aResult['foundorder'] += 0.0001 * $aClassType[$aResult['class'].':'.$aResult['type']]['importance'];
1313                 } else {
1314                     $aResult['foundorder'] += 0.01;
1315                 }
1316             }
1317             if (CONST_Debug) var_dump($aResult);
1318             $aSearchResults[$iResNum] = $aResult;
1319         }
1320         uasort($aSearchResults, 'byImportance');
1321
1322         $aOSMIDDone = array();
1323         $aClassTypeNameDone = array();
1324         $aToFilter = $aSearchResults;
1325         $aSearchResults = array();
1326
1327         $bFirst = true;
1328         foreach ($aToFilter as $iResNum => $aResult) {
1329             $this->aExcludePlaceIDs[$aResult['place_id']] = $aResult['place_id'];
1330             if ($bFirst) {
1331                 $fLat = $aResult['lat'];
1332                 $fLon = $aResult['lon'];
1333                 if (isset($aResult['zoom'])) $iZoom = $aResult['zoom'];
1334                 $bFirst = false;
1335             }
1336             if (!$this->bDeDupe || (!isset($aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']])
1337                 && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']]))
1338             ) {
1339                 $aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']] = true;
1340                 $aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']] = true;
1341                 $aSearchResults[] = $aResult;
1342             }
1343
1344             // Absolute limit on number of results
1345             if (sizeof($aSearchResults) >= $this->iFinalLimit) break;
1346         }
1347
1348         return $aSearchResults;
1349     } // end lookup()
1350 } // end class